Issue of building fpc_trunk and lazarus_trunk x64 for Windows

[DonAlfredo]

FPC rev 45216 is also ok.
Newer FPC revisions seem to cause this issue !
Getting closer.

[PascalDragon]

@PascalDragon

Could you please explain ?
The code-sample shows a very simple way of testing this.
This issue has now been reproduced by some people.
Just build FPC trunk and Lazarus trunk. And build+look at startlazarus.

I want one simple program file that I can use for testing that either halts with an exit code <> 0 if the error occurred or with an exit code = 0 if it didn't. And that doesn't need any special options to compile. Something that I can stuff into the compiler's testsuite once this bug is fixed.

[JiaXing]

I want one simple program file that I can use for testing that either halts with an exit code <> 0 if the error occurred or with an exit code = 0 if it didn't. And that doesn't need any special options to compile. Something that I can stuff into the compiler's testsuite once this bug is fixed.

The fact is we have no idea what's wrong with startlazarus. As other said, they can't reproduce the bug outside of startlazarus. So you have to go straight into startlazarus to find the cause. Perhaps by isolate the problematic code and check the produced asembly?

[PascalDragon]

That's why I'm asking someone to reproduce this with a simple example. I have enough other stuff to work on already.

[zoltanleo]

That's why I'm asking someone to reproduce this with a simple example. I have enough other stuff to work on already.

Hi Sven.

If I understand you correctly, then the problem is easy to reproduce. You must (for Windows) build any trunk compiler after revision 45216, then build the latest version of Lazarus and try to run the IDE using starlazarus.exe. This is true for both x32 and x64.

If necessary, I will give a step-by-step code on how this error is reproduced by me.

[PascalDragon]

That's why I'm asking someone to reproduce this with a simple example. I have enough other stuff to work on already.

Hi Sven.

If I understand you correctly, then the problem is easy to reproduce. You must (for Windows) build any trunk compiler after revision 45216, then build the latest version of Lazarus and try to run the IDE using starlazarus.exe. This is true for both x32 and x64.

If necessary, I will give a step-by-step code on how this error is reproduced by me.

I don't want to reproduce this in startlazarus.exe, I want a self contained example that is a simple Pascal program.

[DonAlfredo]

Sorry about this. I have tried to isolate this issue, but did not succeed.
GetExeExt works everywhere in my tests, except in startlazarus.
I think it is related to the inline changes of commit 45217.

[PascalDragon]

Sorry about this. I have tried to isolate this issue, but did not succeed.
GetExeExt works everywhere in my tests, except in startlazarus.
I think it is related to the inline changes of commit 45217.

If you can show that startlazarus fails with 45217, but succeeds with 45216 you can open a bug report with that information as well.

[marcov]

Sorry about this. I have tried to isolate this issue, but did not succeed.
GetExeExt works everywhere in my tests, except in startlazarus.
I think it is related to the inline changes of commit 45217.

If you can show that startlazarus fails with 45217, but succeeds with 45216 you can open a bug report with that information as well.

Unlikely, since that is a z80 related change. More likely 2-3 revs before when there was some inline related work.

[PascalDragon]

Sorry about this. I have tried to isolate this issue, but did not succeed.
GetExeExt works everywhere in my tests, except in startlazarus.
I think it is related to the inline changes of commit 45217.

If you can show that startlazarus fails with 45217, but succeeds with 45216 you can open a bug report with that information as well.

Unlikely, since that is a z80 related change. More likely 2-3 revs before when there was some inline related work.

But 45217 is related to inlining. If it fails with that, then reverting to 45216 (no matter what that was about) will show whether that was the faulty one or not.

[DonAlfredo]

Bugger filed.
https://bugs.freepascal.org/view.php?id=37034

[PascalDragon]

Thank you. As Florian had already noticed other bugs related to inlining he'll hopefully find that as well.

[zoltanleo]

Bugger filed.
https://bugs.freepascal.org/view.php?id=37034

Good news.   

Guys, thanks everyone for participating in the discussion of this problem.